Housing Demand, Savings Gluts and Current Account *Dynamics

Federal Reserve Bank of Dallas Globalization and Monetary Policy Institute

Working Paper No. 221 http://www.dallasfed.org/assets/documents/institute/wpapers/2015/0221.pdf

Housing Demand, Savings Gluts and Current Account Dynamics*

Pedro Gete

Georgetown University and IE Business School

January 2015 Revised: August 2015

Abstract This paper studies the role of housing markets in explaining recent current account dynamics. I document a strong negative correlation, both across and within countries, between housing and current account dynamics. Then, in a quantitative two-country model without exchange rate driven expenditure switching, I analyze savings glut shocks and three drivers of housing demand (population, loan-to-value and housing price expectations) for which I input their dynamics observed in the OECD economies since the mid 1990s. Housing drivers alone imply counterfactual interest rate dynamics. Savings glut shocks alone cannot account for the housing dynamics. The combination of both types of shocks allows to match the emergence and narrowing of the Global Imbalances and the housing booms and busts. Counterfactuals using the model suggest that, as long as loan-to-values are regulated and housing expectations are not very optimistic, the large global imbalances of the mid-2000s are unlikely to return. JEL codes: F32, G28, R21

* Pedro Gete, Georgetown University, Department of Economics, ICC 580, 37th and O Sts., NW, Washington, DC 20057. 202-687-5582. [email protected]. I am very grateful to Anil Kashyap, Sam Kortum and Monika Piazzesi for their advice and support. I thank Fernando Alvarez, Chuqiao Bi, Tim Bian, Steven Davis, Maris Goldmanis, Veronica Guerrieri, Dirk Krueger, Robert Lucas, John Leahy, Priscilla Man, Hector Perez, Martin Schneider, Hyun Song Shin, Nancy Stokey, Harald Uhlig, anonymous referees and workshop participants at several instittions for helpful comments. Financial support from Banco de España, La Caixa and the University of Chicago is gratefully acknowledged. I also thank the Bank of Spain and the Board of Governors of the Federal Reserve for hospitality while part of this research was undertaken. Christophe Andre and Anthony Murphy were very kind to provide some data. The views in this paper are those of the author and do not necessarily reflect the views of the Federal Reserve Bank of Dallas or the Federal Reserve System.

http://www.dallasfed.org/assets/documents/institute/wpapers/2015/0221.pdf

mailto:[email protected]

1 Introduction

Before the 2007 financial crisis, large current account deficits in the U.S. and other OECD

economies attracted considerable attention from academia and policy-makers. The topic was

referred to as "the Global Imbalances" and the main concerns were threefold: 1) the imbalances

may be reflecting intentional distortions, such as unfair trade practices or exchange rate manip-

ulations (see for example IMF 2007); 2) The imbalances were due to domestic distortions, such

as large public deficits or excessive private savings ("savings gluts") which were in individual

economies’self-interest to correct (Bernanke 2005); 3) The adjustment of the imbalances would

require large exchange rate adjustments (a dramatic dollar depreciation) to induce expenditure

switching from foreign to domestic goods and services (Krugman 2007, Obstfeld and Rogoff

2005). A decade later, the imbalances have narrowed markedly, exchange rate adjustments

have played a limited role, and the open question is if the narrowing is temporary or permanent

(IMF 2014).

In this paper I show that both the expansion and shrinking of the Global Imbalances since

the mid 1990s is intimately connected with the dynamics of housing markets. First, I document

that, during the last two decades, there has been considerable heterogeneity in the dynamics of

housing markets and the current account in OECD economies. These dynamics have a strong

negative correlation, both within and across countries. For example, countries like Spain or the

U.S., among others, had large housing booms and current account deficits. Current account

reversals coincided with a decline in housing markets. Meanwhile, in countries like Canada,

Germany or Switzerland, residential investment and housing prices decreased in the midst of

large current account surpluses. Reductions in these surpluses coincided with an improvement

in housing dynamics.1

Second, I study a quantitative two-country model with housing. The model has no role

for exchange rate adjustments to induce expenditure switching as there is only one tradable

good. The dynamics of the current account are driven by expenditure expansion or reduction.

Increases in the demand for housing lead to a current account deficit for three reasons: 1) Higher

housing prices soften collateral constraints and allow an increase in consumption and imports

that generates a current account deficit; 2) Building houses requires imports of tradable goods

for construction, appliances, furniture, utilities and related sectors; 3) Residential investment

promotes reallocation of labor and capital from industries producing tradable goods towards

nontradable industries as construction. Trade deficits decouple consumption from production,

1For data availability reasons I focus on OECD economies. China conforms to the group of countries whosecurrent account surplus shrank after 2007 while its housing markets boomed.

2

and countries import tradable goods to replace the goods that used to be produced by the

inputs reallocated to building houses.

I analyze the three drivers of housing markets most popular in the real estate literature:

1) Population dynamics. Between 1994 and 2006 immigration to countries like Spain and the

U.S. led to nearly a 20% increase in population. A large body of literature has documented

that immigration pushes up housing values, see for example Otterstrom (2015) or Saiz (2007).

2) Changes in collateral requirements (loan-to-value, LTV). Several authors like Corbae and

Quintin (2015), Favilukis et al. (2010) or Kermani (2012) argue that looser credit standards

helped feed the housing boom, and then their reversal led to the subsequent bust. 3) Changes

in housing price expectations. Cheng et al. (2013), Foote et al. (2012), Garriga et al. (2012),

Gelain et al. (2015), Ling et al. (2013), Soo (2013), and Van der Cruijsen (2014), among others,

provide evidence that homebuyers’beliefs played a key role in recent housing dynamics.

If we input dynamics for population, LTVs and housing price expectations similar to those

observed in OECD economies since the mid 1990s, then the model generates dynamics for

housing quantities and prices (including both prices and price-to-rent ratios), and for the current

account that are similar to the data, both in size and in timing. However, housing demand

drivers alone cannot explain the whole story. The model driven by the three housing drivers

generates counterfactual increases in interest rates because housind demand encourages higher

credit demand. When I add a credit supply shock to the simulations (a foreign savings glut),

then the model can account for housing, current account and interest rate dynamics. Savings

glut shocks alone are not enough to match the observed housing dynamics and would lead to

counterfactual housing price-to-rent ratios.

The previous exercise suggests that housing demand may have been a key driver of current

account dynamics, both during the period of the Global Imbalances, and during the sharp

reversal of the deficits that took place after the housing collapse. Savings glut shocks reinforced

the housing drivers. Exchange rate-induced expenditure switching does not seem very important

to understand recent current account dynamics, as a model abstracting from it is consistent

with the data.

Finally, I use the model to perform counterfactuals in which I simulate current account

dynamics when LTVs are regulated and expectations of housing prices are moderate. These

two assumptions seem to be the new normal for housing markets in most OECD economies.

The exercise suggests that we will not see the reappearance of the Global Imbalances even in

the presence of savings gluts.

This paper complements alternative theories for the Global Imbalances. For example,

3

Backus et al. (2014) consider the role of demographic trends in an OLG model in which

saving decisions are tied to life expectancy. Barattieri (2014) proposes an explanation based

on the U.S. comparative advantage in services and the asymmetric trade liberalization process

in goods trade versus service trade. Broer (2014) studies models in which higher income risk

can explain the observed fall in the U.S. asset position. Caballero et al. (2008) model the

savings glut hypothesis proposed by Bernanke (2005). Eugeni (2015) theorizes the savings

glut in a two-country OLG model with pay-as-you-go pension systems. Fogli and Perri (2006)

show that reductions in aggregate volatility caused by the “Great Moderation”could have re-

duced precautionary savings in the U.S. more than in other countries and caused a current

account deficit. Jacob and Peersman (2013) estimate that shocks to the marginal effi ciency of

investment are the main driver of cyclical fluctuations in the U.S. trade balance. Mendoza et

al. (2009) attribute the current account imbalances to financial globalization among countries

with idiosyncratic risks and heterogeneous domestic financial markets. Housing demand is an

interesting complementary explanation because it can account for the heterogeneity in the cur-

rent account positions of countries with similar levels of financial development. Models focused

on U.S.-specific factors have trouble explaining why the dynamics of the U.S. current account

have been so similar to other developed economies.

There are a few papers that address the link between housing markets and the Global

Imbalances. Aizenman and Jinjarak (2013) and this paper are the first papers to document

and study the strong correlation between housing and current account dynamics both during

the housing boom and bust periods. Aizenman and Jinjarak (2013) is a panel regression study

that shows that the most significant variable in accounting for real estate valuation changes is

the lagged real estate valuation appreciation, followed by lagged declines of the current account

to GDP ratio. Other papers have focused on the correlation between housing and the current

account during the housing boom. Gete (2009) documents the correlation during the booms,

and theorizes that input reallocation between tradable and non-tradable sectors may help in

explaining why housing demand can generate trade deficits. Matsuyama (1990) theoretically

studies the current account consequences of income effects on residential investment. Laibson

and Mollerstrom (2010) relate housing and current account dynamics assuming a behavioral

bubble and aggregate wealth effects. Adam et al. (2011) study a small open economy model

with a collateral constraint in which Bayesian learning about housing prices amplifies the effects

of interest rate cuts. Punzi (2013) studies business cycle simulations of a two-country version

of the Iacoviello (2005) model of housing collateral effects. Also, using a two-country version

of Iacoviello (2005), Ferrero (2013) focuses on impulse responses to monetary policy and LTV

shocks. To maximize the collateral channel, he assumes that all agents in the domestic economy

are constrained. Basco (2014), Justiniano et al. (2014) and Favilukis et al. (2012) study the

4

effects of the Global Imbalances on housing markets. Basco (2014) shows that globalization

makes housing rational bubbles more likely to appear in developed countries and documents

that an increase in the current account deficit raised U.S. housing prices. Favilukis et al. (2012)

argue that changes in international capital flows played at most a small role in driving housing

price movements in the recent years, and that the key causal factor was a financial market

liberalization and its subsequent reversal. Justiniano et al. (2014) claim that foreign capital

flows account for between one fourth and one third of the increase in U.S. housing prices.

The paper proceeds as follows. Section 2 documents three facts about housing and current

account dynamics. Section 3 describes the model and Section 4 the calibration. Section 5

analyzes the main mechanisms that connect housing and current account dynamics. Section 6

discusses the main quantitative exercise and Section 7 the counterfactual experiment. Section

8 concludes.

2 Some Facts about Housing and Current Account Dy-

namics

In this Section I document three facts that motivate the remainder of the paper. First,

as Figure 1 illustrates, several OECD economies have had large and persistent current account

deficits since the mid 1990s. Most current account deficits have decreased significantly since

2006. The U.S. is not a special case; its current account dynamics have been similar to those

of several other countries.

Second, there has been substantial heterogeneity in both the current account and housing

dynamics of developed economies. For example, countries like Spain or the U.S. have had large

increases in residential investment, housing prices and employment in construction since the

mid 1990s to around 2006. Meanwhile, real housing prices and residential investment decreased

in countries like Germany or Switzerland, among others. The dynamics reversed after 2006,

when housing markets collapsed in countries like Spain or the U.S. and started to rise in the

countries that did not experience a boom in the previous decade. The x-axis in Figure 2 shows

the wide heterogeneity in housing dynamics among OECD countries. The y-axis shows the

heterogeneity in the dynamics of the current account to GDP ratio.2

2The Figure contains all OECD countries for which I found available data in the OECD database. For housingprices these countries are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece,Iceland, Ireland, Israel, Italy, Japan, South Korea, Luxembourg, Netherlands, New Zealand, Portugal, Spain,Sweden, Switzerland, UK and US. For residential investment and employment in construction some countrieswere not available. I excluded Norway because of the weight of oil prices in its current account dynamics.

5

Third, changes in housing dynamics have a strong negative correlation with changes in

current account dynamics, both within and across countries. The strong correlation holds both

during the period of housing booms (mid-1990s to around 2006), and during the period of

housing busts (2007-2012). Figure 2 shows the cross-country correlations. Figure 3 focuses on

the within-country correlations. The left column of Figure 2 contains scatterplots of changes

in housing variables and changes in the current account ratios between 1996 and 2006, while

the right column redoes the scatterplots for the period from 2007 through 2012. Housing

variables and the current account had monotonic behavior between these dates. Countries that

experienced housing booms also had larger current account deficits. Moreover, the current

account reversals coincided with the decline in housing markets.3 The heterogeneity within

Europe is especially interesting, because the European Union as a whole had a nearly balanced

current account.

3 Model

There is a domestic and a foreign country. In both countries, there is a housing sector,

which is non-tradable, and a sector producing tradable goods. All trade between countries is

intertemporal since there is only one tradable good.

3.1 Domestic Households

At period t there is a mass Nd,t of infinitely-lived domestic households who can be patient

or impatient. These two types differ in three dimensions: 1) The discount factor for the patient

households is larger than for the impatient households (βp > βi).4 2) The impatient households

face a collateral constraint that limits their borrowings to a fraction of the discounted expected

value of the houses they hold. 3) Patient domestic households have access to two types of

one-period bonds: an international bond, B̂, with real interest rate R̂, to borrow or save

with the foreign households; and domestic bonds, B, with real interest rate R, to lend to the

domestic impatient households. A non-arbitrage condition governs the relation between the

two types of bonds. The impatient domestic households can only borrow from the domestic

patient households. This is a simplifying assumption without loss of generality. In fact, the

impatient domestic households can borrow from the foreign households through the domestic

3Anecdotal evidence suggests that emerging markets also followed the patterns reported in Figure 3.4This is a standard mechanism to allow for credit relations in which the impatient household borrow from

the patient household (Iacoviello 2005).

6

patient households, who in that regard behave as financial intermediaries.

Households supply labor inelastically in their home country. Every period in the domestic

country, there are (1− φ)Nd,t patient households, and φNd,t impatient households. The para-

meter φ controls both the share of impatient households over the total domestic population,

and their share in the income of the domestic country. The total population of the domestic

country, Nd,t, can change over time to analyze how population dynamics affect housing markets.

3.1.1 Domestic Patient Households

There is a representative domestic patient household that maximizes the expected utility

of its members

E0

∞∑t=0

βtp (1− φ)Nd,tu(cpd,t, h

pd,t

), (1)

where cpd,t and hpd,t are the per capita consumption of tradable goods and housing. The flow

of housing consumption is equal to the per capita stock of housing. Preferences are constant

relative risk aversion over a constant elasticity of substitution aggregator of housing services

and tradable goods consumption

u(cpd,t, h

pd,t

)=

[[(1− θ)

(cpd,t) ε−1

ε + θ(hpd,t) ε−1

ε

] εε−1]1− 1

σ

1− 1σ

, (2)

where σ is the elasticity of intertemporal substitution as well as the inverse of the coeffi cient

of relative risk aversion. ε is the static, or intratemporal, elasticity of substitution between

housing and tradable goods consumption. θ ∈ (0, 1) is a parameter that affects the share ofconsumption of housing services in total expenditure.

The aggregate variables for the domestic patient households are: Cpd,t = (1− φ)Nd,tc

pd,t,

Hpd,t = (1− φ)Nd,th

pd,t, B

pd,t = (1− φ)Nd,tb

pd,t, and B̂d,t = (1− φ)Nd,tb̂d,t. b̂d,t are the patient

households’per capita holdings of the international bond, and bpd,t are the per capita holdings

of domestic bonds.

The budget constraint for the representative domestic patient household is

Cpd,t +Bp

d,t + B̂d,t + qd,t(Hpd,t − (1− δ)H

pd,t−1

)+ (1− φ)Nd,t

ψB2b̂2d,t =

= Rt−1Bpd,t−1 + R̂t−1B̂d,t−1 + (1− φ) Id,t, (3)

7

where qd,t is the price of a domestic house in terms of tradable goods, δ is the house depreciation

rate, Rt is the domestic gross real interest rate, R̂t is the international gross real interest rate,

Id,t is the households’ income (to be defined below), ψB is the parameter that controls the

adjustment costs in the holdings of international bonds. The adjustment costs ensure that

there is a unique steady state (Schmitt-Grohe and Uribe 2003).

The first order conditions of the domestic patient households give the non-arbitrage restric-

tion between the return of the two bonds:

Rt

[1 + ψB b̂d,t

]= R̂t. (4)

Both bonds give the same return when the adjustment cost goes to zero, as well as in the steady

state.

3.1.2 Domestic Impatient Households

The representative domestic impatient household maximizes the expected utility of its

members

E0

∞∑t=0

βtiφNd,tu(cid,t, h

id,t), (5)

u(cid,t, h

id,t

)=

[[(1− θ)

(cid,t) ε−1

ε + θ(hid,t) ε−1

ε

] εε−1]1− 1

σ

1− 1σ

, (6)

where all variables are as defined for the patient household, but now they have the superscript

of the impatient household. I assume that βi < βp. The aggregate variables for the impatient

households are Cid,t = φNd,tc

id,t, H

id,t = φNd,th

id,t and B

id,t = φNd,tb

id,t.

The representative domestic impatient household chooses per capita housing, tradable con-

sumption, and domestic bond holdings(bid,t)to maximize (5) − (6) subject to her aggregate

budget constraint:

Cidt +Bi

dt + qdt(H idt − (1− δ)H i

dt−1)= Rt−1B

id,t−1 + φId,t. (7)

Impatient households’per capita borrowings cannot be larger than a fraction mt of the

8

discounted future value of their current houses. That is,

bidt =−mtEt (qd,t+1h

idt)

Rt

. (8)

3.2 Domestic Firms

Firms produce tradable goods (YT ) using labor (NT ). Tradable goods can be used for

consumption by households in both countries, or as housing appliances (Ya). New houses (Yh)

are produced using non-tradable housing structures (Ys), and tradable goods to which I refer

as the housing appliances. The housing structures are built using labor (Ns) and land (L) . The

production functions are:

YTd,t = NαTd,t, (9)

Ysd,t =[Nαsd,t

]γL1−γd , (10)

Yhd,t = min (Ysd,t, τYad,t) , (11)

where α, γ, τ are parameters. Every period there is an exogenous flow of land L. The subscript

d denotes domestic variables. The Leontief assumption in (11) captures the complementarities

between tradable and non-tradable goods in producing houses. It implies that, in equilibrium,

Ysd,t = τYad,t.

There is a quadratic adjustment cost (ψn) to moving labor across sectors. The cost is

paid in units of tradable goods. Since the domestic households own the firm and the land,

households’income is the firms’revenue from selling new houses and new tradable goods net

of the appliances used to produce houses and the adjustment costs:

Id,t = qd,tYhd,t + YTd,t − Yad,t −ψn2(Nsd,t −Nsd,t−1)

2 . (12)

3.3 Foreign Country

To simplify, I assume there are only patient unconstrained households in the foreign coun-

try. The representative foreign household chooses per capita consumption of tradable goods,

9

non-tradable foreign housing, and international bonds(b̂f,t

)to maximize

E0

∞∑t=0

βtpNf,tu(cf,t, hf,t), (13)

u(cf,t, hf,t) =

[[(1− θ)c

ε−1ε

f,t + θhε−1ε

f,t

] εε−1]1− 1

σ

1− 1σ

. (14)

subject to her aggregate budget constraint:

Cf,t + B̂f,t + qf,t (Hf,t − (1− δ)Hf,t−1) +Nf,tψB2b̂2f,t = R̂t−1B̂f,t−1 + If,t. (15)

The aggregate variables for the foreign households are Cf,t = Nf,tcf,t, Hf,t = Nf,thf,t and

B̂f,t = Nf,tb̂f,t.

Foreign firms have the same technology as domestic firms:

YTf,t = NαTf,t, (16)

Ysf,t =[Nαsf,t

]γL1−γf , (17)

Yhf,t = min (Ysf,t, τYaf,t) , (18)

where NTf,t and Nsf,t are the amounts of labor allocated to tradable goods and the housing

sector in the foreign country. The income of foreign households is the total revenue of the

foreign firms:

If,t = qf,tYhf,t + YTf,t − Yaf,t −ψn2(Nsf,t −Nsf,t−1)

2 . (19)

3.4 Market Clearing

In each country, the labor used to produce in the two sectors must equal the total labor

supply:

NTd,t +Nsd,t = Nd,t, (20)

NTf,t +Nsf,t = Nf,t. (21)

10

The increase in the housing stock is the new houses produced minus the depreciation:

H id,t +Hp

d,t − (1− δ)(H id,t−1 +Hp

d,t−1)= Yhd,t, (22)

Hf,t − (1− δ)Hf,t−1 = Yhf,t. (23)

Tradable goods are used for consumption, as housing appliances, and to pay for the portfolio

and labor movement adjustment costs:

Cft + Cpd,t + Ci

d,t + Yad,t + Yaf,t + (1− φ)Nd,tψB2b̂2d,t +Nft

ψB2b̂2f,t

= YTd,t + YTf,t −ψN2(Nsd,t −Nsd,t−1)

2 − ψN2(Nsf,t −Nsf,t−1)

2

The net supply of domestic bonds between the patient and impatient households equals

zero:

Bpd,t +Bi

d,t = 0. (24)

The net supply of international bonds between the two countries equals zero:

B̂d,t + B̂f,t = 0. (25)

The trade balance is the difference between the tradable goods produced and those con-

sumed:

TBd,t = YTd,t − Yad,t − Cpd,t − Ci

d,t − (1− φ)Nd,tψB2

(b̂d,t

)2− ψn2(Nsd,t −Nsd,t−1)

2 .

While the current account is the change in the net foreign asset position:

CAd,t = B̂d,t − B̂d,t−1. (26)

4 Calibration

I calibrate the model using aggregate and micro data from OECD countries, although for

some series only U.S. data were available. Some parameters are exogenously selected based

on values that are common in the literature, or on micro-evidence. The other parameters are

selected for the steady state of the model to match some key statistics. In the steady state

there is no international debt(B̂d = 0

). I assume that one period in the model is one year.

11

1. Exogenously selected parameters. For the intertemporal elasticity of substitution (σ) ,

I follow the real business cycle literature that usually assumes σ = 12, which under CRRA

preferences implies a value for risk aversion of 2. Concerning the elasticity of substitution

between consumption of goods and housing services, several papers have argued for elasticities

below 1, implying complementarity between tradable goods and housing services. For example,

Davidoff and Yoshida (2008) obtain estimates for this elasticity ranging from 0.4 to 0.9. Kahn

(2008) provides evidence based on both aggregate and microeconomic data that is less than one.

Lustig and Van Nieuwerburgh (2006) use 0.05 to match the volatility of U.S. rental prices in an

asset pricing model with housing collateral. Flavin and Nakagawa (2002) estimate 0.13 between

housing and nondurable consumption (proxied by food consumption at home and eaten out).

Since a key element of housing in my model is its nontradability, I work with ε = 0.4, a value

close to the estimate in Tesar (1993) that the elasticity between traded and nontraded goods

is 0.44.

I assume the same labor share across sectors and set it to the standard α = 0.67. For the

depreciation of the stock of houses, I use 2% annual depreciation, δ = 0.02, which is consistent

with the report from the Bureau of Economic Analysis (2004) that annual depreciation rates

for one-to-four-unit residential structures are between 1.1% and 3.6%.

2. Endogenously selected parameters. I set the discount factor of the patient households to

βp = 0.97 to target a 3% annual real interest rate in the steady state. As discussed in Iacoviello

and Neri (2010), the impatient households’discount factor (βi) needs to be small enough to

guarantee that the borrowing constraint (8) is always binding. For an annual model, I choose

βi = 0.85, which is within the range of values used in the literature. For example, in quarterly

models, Iacoviello (2005) chooses βi = 0.95 while Punzi (2013) uses βi = 0.98. Ferrero (2014)

argues that the choice of βi depends on the change in the LTV ratio. In a quarterly model, he

chooses βi = 0.96 when the LTV changes from 0.75 to 0.99, and a smaller βi = 0.89, when the

LTV changes from 0.85 to 0.95.

There is no consensus in the literature regarding the share of households whose borrowing is

constrained. This is an important parameter for the reaction of the domestic economy to LTV

shocks. In the standard life-cycle model with one risk-free asset, the fraction of constrained

households is very small (usually below 10%) under parameterizations where the model’s dis-

tribution of net worth is in line with the data (Heathcote et al. 2009). On the other extreme,

Ferrero (2014) assumes that 100% of households face borrowing constraints. Iacoviello (2005)

estimates that 64% of the wage income goes to the patient households. I assume that 40%

of the domestic households are impatient (φ = 0.4). This number is consistent with recent

papers which measure the share of constrained households using data on liquidity constrained

12

households. For example, Justiniano et al. (2014), using different U.S. Surveys of Consumer

Finances (SCF), estimate that these households represent 61% of the population and 46% of

the labor income. Kaplan and Violante (2014) find that between 25% and 66% of households

hold sizeable amounts of illiquid wealth, yet consume all of their disposable income during a

pay-period. Lusardi et al. (2011) show that 25% of U.S. households are certainly unable to

"come up with $2,000 within a month", and 49% probably could not come up with the $2,000

at all.

I choose the steady state value of the LTV parameter, m = 0.92, to match the 1994 me-

dian LTV for first-time home buyers (the most important marginal group of home buyers), as

computed by Duca et al. (2011). I normalize population to be one in the steady state. The

remaining six parameters (τ, θ, γ, ψn, ψB,LdNd) control the size of the housing sector, appliances

and the elasticity of the housing supply. I calibrate them to match the following six targets in

a world with symmetric country sizes in the steady state:5 1) A ratio of residential investment

to output of 5%. This is the U.S. long-term average. 2) A ratio of spending on housing services

relative to consumption of durables and services of 17% (Davis and Van Nieuwerburgh 2014).

The level of housing costs in household budgets varies from 16% to 27% in the OECD countries

(OECD 2011). 3) The average homebuyer spends around 5% of the value of their house on ap-

pliances, furnishings, and remodeling activities (Siniavskaia 2008). 4) The share of employment

in the construction sector is 5% (Boldrin et al. 2013). 5) The aggregate housing price-to-rent

ratio is 22 (Davis et al. 2008). 6) An average price-elasticity of housing supply equal to 1.15

over the first two years. This value is consistent with the evidence for OECD economies of

Caldera and Johansson (2013). The parameters of the calibrated model are summarized in

Table 1.

5 Impulse Responses

To illustrate how the model connects housing and current account dynamics, this Section

analyzes impulse responses for different values of the parameters. First, Figure 4 reports the

responses to an increase in the expectations of domestic housing prices for high and low housing

supply elasticity. When the land share in structures is high (γ is low) housing supply is inelastic

since most of the structures are land, which is exogenously fixed. All panels of Figure 4 assume

that the share of impatient households is zero (φ = 0) to shut down the collateral consumption

channel which is analyzed in Figure 5. Figure 4 shows that the current account reacts more

when housing supply is elastic. This is because construction needs tradable goods (housing ap-

5That is, Nd = Nf , Ld = Lf .

13

pliances), and because the reallocation of labor towards nontradables (construction) encourages

imports of consumption goods to smooth the opportunity cost of building new houses, which

is the foregone production of tradable goods. By importing consumer tradables the economy

can build non-tradables while still consuming tradables.6

Figure 5 reports the responses to the same shock than in Figure 4 but alters the share of

impatient households (φ). The current account reacts much more when the share of impatient

households is large. Higher expected housing prices leads to higher current prices, more collat-

eral value of housing and larger borrowing by the constrained households. These households,

given their low discount factor, allocate most of their new borrowing into consumption of the

non-durable good, which is tradable (Figure 5c). This accounts for the larger current account

deficit of Figure 5d. However, as Figure 5b plots, if the share of impatient households is large

enough the collateral mechanism leads to a counterfactual observation: the housing price-to-

rent ratio decreases.7 Housing prices (the value of the housing asset) increase less than housing

rents (the value of the housing flow) because the collateral channel encourages the consumption

by the impatient households, who value less the durable good.

6 Simulations

This Section simulates the model for three sets of exogenous shocks. One set has only

the housing demand drivers: population, LTV and housing price expectations. In the figures

I refer to this set of shocks as "Model, housing drivers". The second set only has a shock to

the foreign discount factor ("Model, savings glut" in the figures). The third set of shocks is

the combination of the housing and savings glut shocks (I refer to this set as "Model, housing

drivers + savings glut"). Section 6.1 discusses how I use data to discipline the exogenous shocks.

Then I input the exogenous shocks into the model and report the reactions of its endogenous

variables comparing them with OECD data.8 This is the same methodology that Garriga et

al. (2012) and Justiniano et al. (2014) use to analyze U.S. housing markets, and how Meza

and Urrutia (2011) study exchange rates and net exports dynamics. The goal is to evaluate

6Gete (2009) analyzes theoretically this mechanism.7The relation between the rental and house price is

qt = pl,t + β(1− δ)Et[qt+1

uc,t+1uc,t

],

where pl,t is the rental rate and uc,t the marginal utility of consumption.8The model is solved using a nonlinear Newton-type algorithm (Adjemian et al. 2011) for a perfect foresight

version.

14

the ability of the model and its driving forces to account for both housing and current account

dynamics.

6.1 Driving Forces

As driving forces I use savings glut shocks and three drivers of housing markets:

1) Housing price expectations that I measure from survey data collected by Case et al.

(2012). They surveyed around 5000 recent homebuyers in four U.S. counties regarding the

nominal housing prices they expected to see next year.9 To construct series of expectations of

real prices I merge the Case et al. (2012) data with the inflation expectations from the Michigan

Survey of Consumers. The top panel of Figure 6 compares the expectations of real housing

prices (dashed lines) for each county with the realized housing prices (solid lines).10 The bottom

panel of Figure 6 redoes the top panel but for the expectations of housing price growth. Figure

6 shows that households underestimated housing price growth until 2005. When I give shocks

to housing price expectations I impose them to generate expectations close to those reported

in Figure 6.11 This can be seen in Figure 7c, which contains the exact expectations that the

model uses and also the data from Figure 6. In the case with only the savings glut shock the

housing expectations are fully rational and totally endogenous (model line denoted as "Model,

savings glut").

2) The model line in Figure 7a plots the dynamics of population used for the model simu-

lations. It is close to the experiences of Spain and the U.S. In these countries immigration led

to nearly a 20% increase in population between 1994 and 2006.

3) Figure 7b plots the median LTV series for first time home buyers estimated by Duca et

al. (2011) and the series I feed into the model (variable mt). The U.S. data estimated by Duca

et al. (2011) clearly show an increase in loan-to-values from the mid-1990s until 2006, at which

point a reversal occurred. I could not find an equivalent series for more countries, but anecdotal

9To my knowledge, Case et al. (2012) is the longest survey with quantitative data on expected housingprice growth. The data start in 2003. Table 41 in the Michigan Survey, which has been available since 1978,offers qualitative answers to the question of when is a good time to buy a house. To interpolate the series ofexpectations back to 1994, I used the average growth of real expected house prices computed with the Caseet al. (2012) data for 2003-2006. The series are consistent with the qualitative answers from Table 41 of theMichigan Survey.10I computed the realized prices using housing prices from Freddie Mac and inflation from the Bureau of

Labor Statistics.11Technically, when I input price expectations in the Euler equations I replace qd,t+1 by an expected price

qed,t+1 = qd,t+1 + et. Then, I input a series of et shocks such that qed,t+1 matches the data from Case et al.(2012). In steady state there are no expectation shocks and expectations match realized house prices. Garrigaet al. (2012) use a similar methodology to give shocks to expectations.

15

evidence suggests LTV ratios were relaxed in many other countries. For example, Akin et al.

(2014) document how the manipulation on appraisal values permitted Spanish banks to lend

at higher LTV ratios than what banking regulations allowed. I assume that the LTV returns

to the steady state in about 30 years. The results that I obtain for the housing boom are very

similar to Justiniano et al. (2014), who also matched Duca’s LTV series up to 2006 but then

assumed that the agents take the 2006 LTV levels as permanent.

4) Figure 7d shows that the housing drivers alone generate counterfactual dynamics for

interest rates. Higher housing demand increases credit demand and interest rates rise to achieve

equilibrium. The foreign savings glut is a credit supply shock which can generate the right

path for interest rates. Higher demand for savings from the foreigners leads to lower rates.

To discipline the foreign discount factor shocks I impose that the model generates real interest

rates with a downward trend as in Figure 7d.

6.2 Endogenous Dynamics

Figure 8 contains the endogenous reaction of the model when I input the driving forces

discussed above. Panels a, b, c and d contain the reaction of the housing variables in the domes-

tic economy (housing prices, price-to-rent ratios, employment in construction and residential

investment). The model driven by the three housing drivers generates housing dynamics quite

similar, both in terms of the size of the changes; and in the turning points, to the data from

the countries. The dynamics of the current account are also consistent with the data. However,

the three housing drivers fail to generate decreases in real interest rates (Figure 7d).

The model driven only by savings glut shocks replicates the dynamics of real interest rates

but fails to generate the right comovement between housing prices and price-to-rent ratios.

Impatient households are the most sensitive to interest rate changes and they value more the

flow of housing (rental rate) than the stock (housing prices). Garriga et al. (2012) pointed

out a similar problem: when their perfect foresight model generates decreasing interest rates

it cannot explain both the dynamics of the price-to-rent ratio and of housing prices.12 Here

the model can better match the data because the expected increases in housing prices from the

Case et al. (2012) survey lead to an increase in the asset value of housing.

Combining housing drivers and savings glut shocks allow to account for both housing and

interest rate dynamics. Higher expected prices encourage demand for homeownership, not for

12Shocks to the preferences for housing, which drive housing dynamics in most of the macro-housing literature,would also generate counterfactual price-to-rent ratios and interest rates because they increase the preferencefor the housing flow.

16

rental. Foreign demand for savings lowers interest rates.

Panels e and f of Figure 8 report the domestic and foreign current account dynamics. Like

in the data, the countries with an increase in housing prices and residential investment run a

current account deficit. Increases in housing prices soften collateral constraints, the constrained

households borrow more and allocate most of their borrowings to consumption of tradable

goods, thus pushing the current account towards a deficit. Moreover, the construction sector

imports tradable goods as housing appliances or furniture. The foreign economy runs a current

account surplus while lending to the housing booming country.

The reversal of the current account in the domestic economy is driven by the collapse of

the housing boom. Lower housing prices tighten collateral constraints and reverse the imports

for consumption. Moreover, activity in the construction sector slows with the collapse of em-

ployment in construction after 2007 (Panels c and d in Figure 8). Once the housing boom is

gone in the domestic economy, the foreign economy starts to run a current account deficit, and

housing prices and residential investment increase. These dynamics are very similar to those of

countries like Canada.

7 Counterfactuals

Following the recent financial crisis, many countries have imposed regulations capping LTV

around 80% (Claessens et al. 2014). Moreover, it does not seem likely that in the short-run

households’expectations of housing prices display the optimism of the housing boom period.

Figure 9 analyzes the implications of this new environment for current account dynamics. It

compares the combination of the housing and savings glut shocks that provided the best match

to the data in Figure 8 ("Model, housing drivers + savings glut"), with two counterfactuals.

In both counterfactuals LTV is fixed at 80% (Figure 9a) and the savings glut and population

dynamics are as in Figure 8. However, in one case housing prices are expected to increase by

2% annually, while in the other case housing prices are expected to decrease 5% in the short

run and then later on increase by 1% annually.

Figure 9 shows that, even if the savings glut persists, lower housing demand translates into

lower interest rates, lower housing dynamics and a smaller current account deficit. Thus, the

counterfactuals suggest that it is unlikely to have a comeback of the Global Imbalances as long

as housing markets do not repeat the dynamics of the decade before the financial crisis.

17

8 Conclusions

This paper documented a strong correlation, both across and within countries, between

housing and current account dynamics over the decade before the 2007 financial crisis and

also in the years after it. Then, using a quantitative two-country model, I showed that the

combination of housing demand drivers and savings glut shocks can generate housing booms

and busts together with the emergence and contraction of large current account deficits. The

dynamics are similar to the OECD data, also for the interest rate. The model does not use

exchange rate driven expenditure switching to account for the data. Counterfactuals using the

model suggest that the large Global Imbalances of the mid-2000s are a past phenomenon unlikely

to return as long as LTVs are regulated and housing expectations are not very optimistic.

References

Adam, K., Kuang, P. and Marcet, A.: 2011, House Price Booms and the Current Account,

NBER Chapters pp. 77—122.

Adjemian, S., Bastani, H., Juillard, M., Mihoubi, F., Perendia, G., Ratto, M. and Villemot, S.:

2011, Dynare: Reference Manual, version 4, Dynare Working Papers, Vol. 1, CEPREMAP.

Aizenman, J. and Jinjarak, Y.: 2014, Real Estate Valuation, Current Account and Credit

Growth Patterns, Before and After the 2008-9 Crisis, Journal of International Money and

Finance .

Akin, O., Montalvo, J. G., Villar, J. G., Peydró, J.-L. and Raya, J. M.: 2014, The Real Estate

and Credit Bubble: Evidence from Spain, SERIEs 5(2-3), 223—243.

Backus, D., Cooley, T. and Henriksen, E.: 2014, Demography and Low-Frequency Capital

Flows, Journal of International Economics 92, S94—S102.

Barattieri, A.: 2014, Comparative Advantage, Service Trade, and Global Imbalances, Journal

of International Economics 92(1), 1—13.

Basco, S.: 2014, Globalization and Financial Development: A Model of the Dot-Com and the

Housing Bubbles, Journal of International Economics 92(1), 78—94.

Bernanke, B.: 2005, The Global Saving Glut and the US Current Account Deficit, Sandridge

Lecture .

18

Boldrin, M., Garriga, C., Peralta-Alva, A. and Sánchez, J. M.: 2013, Reconstructing the Great

Recession, Federal Reserve Bank of St. Louis Working Paper Series (2013-006).

Broer, T.: 2014, Domestic or Global Imbalances? Rising Income Risk and the Fall in the US

Current Account, Journal of Monetary Economics 64, 47—67.

Bureau of Economic Analysis: 2004, BEA Depreciation Estimates.

Caballero, R. J., Farhi, E. and Gourinchas, P.-O.: 2008, An Equilibrium Model of Global

Imbalances and Low Interest Rates, The American Economic Review pp. 358—393.

Caldera, A. and Johansson, A.: 2013, The Price Responsiveness of Housing Supply in OECD

Countries, Journal of Housing Economics 22(3), 231—249.

Case, K. E., Shiller, R. J. and Thompson, A.: 2012, What Have They Been Thinking? Home

Buyer Behavior in Hot and Cold Markets.

Cheng, H., Raina, S. and Xiong, W.: 2013, Wall Street and the Housing Bubble.

Claessens, S., Ghosh, S. R. andMihet, R.: 2014, Macro Prudential Policies to Mitigate Financial

Vulnerabilities in Emerging Markets, International Monetary Fund (WP/14/155).

Corbae, D. and Quintin, E.: 2014, Leverage and the Foreclosure Crisis, Journal of Political

Economy .

Davidoff, T. and Yoshida, J.: 2008, Reconciling Micro and Macro Estimates of Substitution

Between Housing and Non-Housing Consumption by Relaxing Homotheticity Restriction.

Davis, M. A., Lehnert, A. and Martin, R. F.: 2008, The Rent-Price Ratio for the Aggregate

Stock of Owner-Occupied Housing, Review of Income and Wealth 54(2), 279—284.

Davis, M. A. and Van Nieuwerburgh, S.: 2014, Housing, Finance, and the Macroeconomy.

Duca, J. V., Muellbauer, J. and Murphy, A.: 2011, House Prices and Credit Constraints:

Making Sense of the US Experience, The Economic Journal 121(552), 533—551.

Eugeni, S.: 2015, An OLG Model of Global Imbalances, Journal of International Economics

95(1), 83—97.

Favilukis, J., Kohn, D., Ludvigson, S. C. and Van Nieuwerburgh, S.: 2012, International Capital

Flows and House Prices: Theory and Evidence, pp. 235—299.

Favilukis, J., Ludvigson, S. and Van Nieuwerburgh, S.: 2013, The Macroeconomic Effects of

Housing Wealth, Housing Finance, and Limited Risk-Sharing in General Equilibrium.

19

Ferrero, A.: 2013, House Price Booms, Current Account Deficits, and Low Interest Rates,

Journal of Money, Credit and Banking .

Flavin, M. and Yamashita, T.: 2002, Owner-Occupied Housing and the Composition of the

Household Portfolio, American Economic Review pp. 345—362.

Fogli, A. and Perri, F.: 2006, The Great Moderation and the US External Imbalance.

Foote, C. L., Gerardi, K. S. and Willen, P. S.: 2012, Why Did so Many People Make so Many

Ex-Post Bad Decisions? The Causes of the Foreclosure Crisis.

Garriga, C., Manuelli, R. E. and Peralta-Alva, A.: 2012, AModel of Price Swings in the Housing

Market, Federal Reserve Bank of St. Louis Working Paper Series (2012-022).

Gelain, P., Bank, N., Lansing, K. J., San Francisco, F. and Natvik, G. J.: 2015, Explaining the

Boom-Bust Cycle in the US Housing Market: a Reverse-Engineering Approach, Federal

Reserve Bank of San Francisco Working Paper Series 2.

Gete, P.: 2009, Housing Markets and Current Account Dynamics, PhD Dissertation, University

of Chicago .

Heathcote, J., Storesletten, K. and Violante, G. L.: 2009, Quantitative Macroeconomics with

Heterogeneous Households, Annual Review of Economics 1(1), 319—354.

Iacoviello, M.: 2005, House Prices, Borrowing Constraints, and Monetary Policy in the Business

Cycle, American Economic Review (95 (3)), 739—764.

Iacoviello, M. and Neri, S.: 2010, Housing Market Spillovers: Evidence from an Estimated

DSGE Model, American Economic Journal: Macroeconomics pp. 125—164.

IMF: 2007, Staff Report on the Multilateral Consultation on Global Imbalances with China,

the Euro Area, Japan, Saudi Arabia and the United States, Washington: International

Monetary Fund .

IMF: 2014, World Economic Outlook October, Washington: International Monetary Fund .

Jacob, P. and Peersman, G.: 2013, Dissecting the Dynamics of the US Trade Balance in an

Estimated Equilibrium Model, Journal of International Economics 90(2), 302—315.

Justiniano, A., Primiceri, G. E. and Tambalotti, A.: 2014, Household Leveraging and Delever-

aging.

20

Justiniano, A., Primiceri, G. E. and Tambalotti, A.: 2014b, The Effects of the Saving and

Banking Glut on the US economy, Journal of International Economics 92, S52—S67.

Kahn, J. A.: 2008, What Drives Housing Prices?

Kaplan, G. and Violante, G. L.: 2014, AModel of the Consumption Response to Fiscal Stimulus

Payments.

Kermani, A.: 2012, Cheap Credit, Collateral and the Boom-Bust Cycle, University of

California-Berkeley, Working Paper .

Krugman, P.: 2007, Will There be a Dollar Crisis?, Economic Policy 22(51), 436—467.

Laibson, D. and Mollerstrom, J.: 2010, Capital Flows, Consumption Booms and Asset Bub-

bles: A Behavioural Alternative to the Savings Glut Hypothesis, The Economic Journal

120(544), 354—374.

Ling, D. C., Ooi, J. and Le, T.: 2013, Explaining House Price Dynamics: Isolating the Role of

Non-Fundamentals.

Lusardi, A., Schneider, D. J. and Tufano, P.: 2011, Financially Fragile Households: Evidence

and Implications.

Lustig, H. and Van Nieuwerburgh, S.: 2006, Can Housing Collateral Explain Long-Run Swings

in Asset Returns?

Matsuyama, K.: 1990, Residential Investment and the Current Account, Journal of Interna-

tional Economics (28 (1)), 137—153.

Mendoza, E., Quadrini, V. and Rios-Rull, V.: 2009, Financial Integration, Financial Deepness

and Global Imbalances, Journal of Political Economy 117(3).

Meza, F. and Urrutia, C.: 2011, Financial Liberalization, Structural Change, and Real Ex-

change Rate Appreciations, Journal of International Economics 85(2), 317—328.

Obstfeld, M. and Rogoff, K. S.: 2005, Global Current Account Imbalances and Exchange Rate

Adjustments, Brookings papers on economic activity 2005(1), 67—146.

OECD: 2011, Housing Conditions, How’s Life?: Measuring Well-being, OECD Publishing .

Otterstrom, S. M. et al.: 2015, Income migration and home price trajectories in the united

states, International Real Estate Review 18(2), 277—302.

21

Punzi, M. T.: 2013, Housing Market and Current Account Imbalances in The International

Economy, Review of International Economics 21(4), 601—613.

Saiz, A.: 2007, Immigration and Housing Rents in American Cities, Journal of Urban Economics

61(2), 345—371.

Schmitt-Grohé, S. and Uribe, M.: 2003, Closing Small Open Economy Models, Journal of

international Economics 61(1), 163—185.

Siniavskaia, N.: 2008, Spending Patterns of Home Buyers.

Soo, C. K.: 2013, Quantifying Animal Spirits: News Media and Sentiment in the Housing

Market.

Tesar, L.: 1993, International Risk-sharing and Non-traded Goods, Journal of International

Economics (35 (1-2)), 69—89.

Van der Cruijsen, C., Jansen, D.-J. and van Rooij, M.: 2014, The Rose-Colored Glasses of

Homeowners.

22

Tables

Table 1: Parameters

Description Parameter Value

Patient households’discount factor βp 0.97

Impatient households’discount factor βi 0.85

Share of impatient households φ 0.4

Intertemporal elasticity of substitution σ 0.5

Intratemporal elasticity of substitution ε 0.4

Housing depreciation rate δ 0.02

Ratio of housing appliances over structures 1τ

0.2

LTV parameter m 0.92

Share of housing services in utility θ 0.18

Labor share in production α 0.67

Land share in housing production 1− γ 0.2

Steady state population Nd = Nf 1

Land supply per capita LdNd=

LfNf

10−5

Labor adjustment cost ψn 7

Adjustment cost on international bond ψB 0.045

23

Figures

1985 1990 1995 2000 2005 201016

14

12

10

8

6

4

2

0

2

4CA/GDP in OECD Countries

Year

Perc

ent

USFranceSpainItalyPortugalGreece

Figure 1. Ratio of the Current Account to GDP for some OECD Countries.

24

50 0 50 100 150 20012

10

8

6

4

2

0

2

4

6

8

% Change Employment in Construction

Chan

ge in

rat

io CA

/GDP

(per

cent

age

point

s)

DE

IT

DK

ES

FI

IR

NLSE

AT

PTGRC

LUX

19962006

Regression Coeff = 0.07tstat = 2.65Rsquare = 0.41

70 60 50 40 30 20 10 0 10 206

4

2

0

2

4

6

8

10

% Change Employment in Construction

Cha

nge

in ra

tio C

A/G

DP

(per

cent

age

poin

ts)

DE

IT

DK

ES

FI

IR

NL

SE

AT

PT

BG

GRC

20072012


40 20 0 20 40 60 80 100 120 140 1608

6

4

2

0

2

4

6

8

10

12

% Change Residential Investment

Chan

ge in

rat

io CA

/GDP

(per

cent

age

point

s)

US

JP

DE

FR

IT

UK

CA

AU

DK

FI

IR

KO

NL

NZ

SE

CH

AT

PT

BG

GRC

ISR

LUX

19962006

Regression Coeff = 0.04

tstat = 2.12

Rsquare = 0.18

80 60 40 20 0 20 40 60 80

8

6

4

2

0

2

4

6

8

10

% Change Residential Investment

Cha

nge

in ra

tio C

A/G

DP

(per

cent

age

poin

ts)

US

JP

DEFR

IT

UK

CA

AU

DK

ES

FI

IR

KO NLNZ

SE

CH

AT

PT

BG

GRC

ISR

20072012


50 0 50 100 150 200 25010

5

0

5

10

15

% Change Real Housing Prices

Chan

ge in

rat

io CA

/GDP

(per

cent

age

point

s)

US

JP

DE

FR

IT

UK

CA

AU

DK

ES

FI

IR

KONL

NZ

SE

CH

PT

BG

ISR

19962006


50 40 30 20 10 0 10 20 30 40 50

6

4

2

0

2

4

6

8

10

% Change Real Housing Prices

Cha

nge

in ra

tio C

A/G

DP

(per

cent

age

poin

ts)

US

JP

DEFR

IT

UK

CA

AU

DK

ES

FI

IR

KONL

NZ

SE

CH

AT

PT

BG

GRC

ISR

20072012


Figure 2. Cross-Country Correlations between Changes in the Current Accountto GDP ratio and Changes in Housing Variables. The first row is the scatter-plot of

the change in the current account to GDP ratio against the change in the share of employment in

construction. The second and third rows replace the x-axis with the change in residential investment

and with the change in housing prices, respectively. The left column shows the 1996-2006 period,

while the right column displays the 2007-2012 period. Data source: OECD.

25

1994 1996 1998 2000 2002 2004 2006 2008 2010 20127

6

5

4

3

2

1%

1994 1996 1998 2000 2002 2004 2006 2008 2010 2012100

110

120

130

140

150

160

Inde

x (1

994

= 10

0)

US

corr(CA,Eh)= 0.93corr(CA,Ph)= 0.91

1994 1996 1998 2000 2002 2004 2006 2008 2010 201215

10

5

0

5

%

1994 1996 1998 2000 2002 2004 2006 2008 2010 201250

100

150

200

250

Inde

x (1

994

= 10

0)

Spain

corr(CA,Eh)= 0.9

corr(CA,Ph)= 0.87

1994 1996 1998 2000 2002 2004 2006 2008 2010 20123

0

3

5

%

1994 1996 1998 2000 2002 2004 2006 2008 2010 201270

100

120

140

180

200

Inde

x (1

994

= 10

0)

France

corr(CA,Eh)= 0.93

corr(CA,Ph)= 0.9

1994 1996 1998 2000 2002 2004 2006 2008 2010 20123

0

3

6

10

%

1994 1996 1998 2000 2002 2004 2006 2008 2010 201260

70

80

90

100

110

Inde

x (1

994

= 10

0)

Germany


1994 1996 1998 2000 2002 2004 2006 2008 2010 20125

3

0

3

5

%

1994 1996 1998 2000 2002 2004 2006 2008 2010 201270

90

100

120

140

Inde

x (1

994

= 10

0)

Italy


1994 1996 1998 2000 2002 2004 2006 2008 2010 20124

2

0

2

4

%

1994 1996 1998 2000 2002 2004 2006 2008 2010 201280

110

140

170

200

Inde

x (1

994

= 10

0)

Canada


Figure 3. Within-Country Correlations between the Current Account (CA),Employment in Construction (Eh) and Housing Prices (Ph). Each panel shows the

dynamics of the current account to GDP ratio (dashed line with scale in the left axis), employment in

construction (dotted line with scale in the right axis) and housing prices (solid line with scale in the

right axis) in an OECD country. The correlations are also displayed. Data source: OECD.

26

0 5 10 15 20 25 30100

100.2

100.4

100.6

100.8

101

101.2

101.4a) Shock to Expected House Prices

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

Number of Periods0 5 10 15 20 25 30

98

100

102

104

106

108

110b) New Houses Built

Number of Periods

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

γ = 0.95, φ=0γ = 0.3, φ=0

0 5 10 15 20 25 3098

100

102

104

106

108

110

112

114

116c) Employment in Housing

Number of Periods

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

γ = 0.95, φ=0γ = 0.3, φ=0

0 5 10 15 20 25 300.05

0.04

0.03

0.02

0.01

0

0.01

0.02d) CA/GDP

Number of Periods

%

γ = 0.95, φ=0γ = 0.3, φ=0

Figure 4. Responses for different housing supply elasticities. These panels compareimpulse responses to an increase in the expectations of domestic housing prices when the supply of

new structures is elastic (low land share, high γ) and when it is not (low γ). All panels are for the

domestic economy. In all panels there are no impatient households (φ = 0) to shut down the collateral

channel.

27

0 5 10 15 20 25 30100

100.2

100.4

100.6

100.8

101

101.2

101.4a) Shock to Expected Housing Prices

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

Number of Periods0 5 10 15 20 25 30

85

90

95

100

105

110

115b) Housing PricetoRent Ratio

Number of Periods

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

φ=0.9φ=0.1

0 5 10 15 20 25 3098.5

99

99.5

100

100.5

101

101.5

102c) Aggregate Consumption

Number of Periods

Inde

x (v

alue

= 1

00 a

t ste

ady

stat

e)

φ=0.9φ=0.1

0 5 10 15 20 25 302

1.5

1

0.5

0

0.5

1d) CA/GDP

Number of Periods

%

φ=0.9φ=0.1

Figure 5. Comparative statics on population share of impatient households. Thisfigure compares impulse responses to an increase in the expectations of domestic housing prices when

the share of the population composed by impatient households (φ) is high or low. All panels are for

the domestic economy.

28

1994 1996 1998 2000 2002 2004 2006 2008 2010 201280

100

120

140

160

180

200

220

240Expected Price vs Realized Housing Prices

Year

Inde

x (y

ear

1994

= 1

00)

Alameda County (realized)Middlesex County (realized)Milwaukee County (realized)Orange County (realized)Alameda County (expected)Middlesex County (expected)Milwaukee County (expected)Orange County (expected)

1996 1998 2000 2002 2004 2006 2008 2010 201215

10

5

0

5

10

15

20

Year

%

Expected minus Realized Housing Price GrowthAlameda CountyMiddlesex CountyMilwaukee CountyOrange County

Figure 6. Comparing expectations from Case et al. (2012) with realized houseprices. The top panel compares the survey data on real house price expectations (dashed lines) fromCase et al. (2012) with the realized real house prices for those counties. The bottom panel redoes the

comparison for the growth rate of house prices.

29

1994 1996 1998 2000 2002 2004 2006 2008 2010 201295

100

105

110

115

120a) Population in Domestic Country

Year

Inde

x (1

994

= 10

0)USUKFranceSpainItalyModel

1994 1996 1998 2000 2002 2004 2006 2008 2010 20120.91

0.92

0.93

0.94

0.95

0.96

0.97

0.98

0.99

1b) LTV in Domestic Country

Year

LTV

valu

e

USModel

1994 1996 1998 2000 2002 2004 2006 2008 2010 201280

100

120

140

160

180

200

220c) Housing Price Expectations in Domestic Country

Year

Inde

x (1

994

= 10

0)

Alameda CountyMiddlesex CountyMilwaukee CountyOrange CountyModel, housing driversModel, savings glutModel, housing drivers + savings glut

1994 1996 1998 2000 2002 2004 2006 2008 2010 20120.94

0.96

0.98

1

1.02

1.04

1.06

1.08d) Domestic Gross Interest Rate

Year

USUKFranceSpainItalyModel, housing driversModel, savings glutModel, housing drivers + savings glut

Figure 7. Driving Forces in Model Simulations. These panels plot the driving forcesof the model. The model dynamics are exogenous in Panels a), b), in c) for the housing drivers and

in d) for the savings glut. The model line "Model, housing drivers" refers to the simulations with the

three housing shocks. The model line "Model, savings glut" refers to the simulations with only the

shock to the foreign discount factor. The model line "Model, housing drivers + savings glut" refers

to the combination of the housing and savings glut shocks. The series are compared with their data

counterparts.

30

1994 1996 1998 2000 2002 2004 2006 2008 2010 201280

100

120

140

160

180

200

220

240

260a) Domestic Housing Prices

Year

Inde

x (1

994

= 10

0)USUKFranceSpainItalyModel, housing driversModel, savings glutModel, housing drivers + savings glut

1994 1996 1998 2000 2002 2004 2006 2008 2010 201260

80

100

120

140

160

180

200

220b) Domestic Housing PricetoRent Ratio

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 201280

100

120

140

160

180

200

220

240c) Domestic Employment in Construction

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 201250

100

150

200

250

300d) Domestic Residential Investment

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 201214

12

10

8

6

4

2

0

2

4

6e) Domestic CA/GDP

Year

Perc

ent


1994 1996 1998 2000 2002 2004 2006 2008 2010 201210

5

0

5

10

15f) Foreign CA/GDP

Year

Perc

ent

CanadaGermanyJapanSwitzerlandModel, housing driversModel, savings glutModel, housing drivers + savings glut

Figure 8. Data vs Endogenous Model Dynamics. This figure compares the endogenousmodel dynamics for the three set of shocks described in Section 6.

31

1994 1996 1998 2000 2002 2004 2006 2008 2010 20120.75

0.8

0.85

0.9

0.95

1a) Domestic LTV

Year

In le

vel

Model, housing drivers + savings glutPessimistic Housing Expectations, LTV regulated2% Expected Increase Housing Prices, LTV regulated

1994 1996 1998 2000 2002 2004 2006 2008 2010 201290

95

100

105

110

115

120

125

130

135b) Domestic Housing Price Expectations

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 20120.98

0.99

1

1.01

1.02

1.03

1.04

1.05c) Domestic Gross Interest Rate

Year

In L

evel


1994 1996 1998 2000 2002 2004 2006 2008 2010 201295

100

105

110

115

120

125

130

135

140

145d) Domestic Residential Inv estment

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 2012100

102

104

106

108

110

112

114e) Domestic Aggregate Consumption

Year

Inde

x (1

994

= 10

0)


1994 1996 1998 2000 2002 2004 2006 2008 2010 201210

8

6

4

2

0

2

4

6

8f) Domestic CA/GDP

Year

%


Figure 9. Counterfactuals. The panels compare the dynamics for the model line "Model,housing drivers + savings glut" of Figure 8 versus two counterfactuals in which LTV is regulated at

80% and expectations of housing prices are either pessimistic or grow at 2%.

32

Housing Demand, Savings Gluts and Current Account *Dynamics

Documents

Transcript of Housing Demand, Savings Gluts and Current Account *Dynamics